@VersunPan Nice! I like to think of my open source projects as plants in my own little garden - slowly improving them when I find time and curiosity to do so. If anyone finds value in anything I share, that's a nice bonus a but not a goal in itself.
Math Major | Daily Diary D0075
Last month I wrote that unsolved questions in mathematics come in two kinds:
🐶 Open Problems - we would recognize a solution the moment we saw one, because! there's an agreed way to check it.
🐱 Open Unknowns - nobody agrees yet on what an answer would even look like.
🐭 In my essay, my main example of an Open Problem was the Riemann Hypothesis. Accordingly, if RH is an OP then we should expect an agreed way to check the proof.
🧀 And there seem to be one - a proof written in Lean can be checked by a computer program, and the check comes out the same for everyone.
🪤 Funny enough, that is exactly what seems to have happened this week. Levent Alpöge - with assistance from Anthhropic's Claude - produced and released a Lean proof of the RH.
⌛️ In less than 24 hours, several independent Lean checkers have accepted it, and it matches the standard statement of RH exactly.
Pretty damn amazing.
🤷♂️ So, what's the issue?
Well, no human-readable proof exists yet, since Alpöge & Claude's proof is over 250,000 pages long.
🦮 It has to be assumed that people who claim to understand the RH proof are referencing an AI-generated substrate and not the contents of the actual 250k pages - which I think would have been the classic interpretation of 'I understand the published proof'.
🌅 So we have a new situation:
a result that is checked - but not understood - at least not in the classical sense.
🚡 But as I wrote in my essay, "a certificate found by any solver ends the question for all of them."
There's no explicit *need* for the certificate to be understood. Being accepted by one single checker is what ends the question.
🦉 Whether a machine check alone counts as "solved" is now the real question. As Terrence Tao seem to suggested: human understandability should remain an important criterion for evaluating AI-generated mathematical proofs.
It will be interesting to see how Dr. Alpöge's historic result affects Professor Tao's standpoint.
🤔 Personally, I think that Alpöge's result is the paradigm shift moment where a new generation of mathematicians abandon the need for "readable and understandable proofs" - and start treating Lean proofs as assembler-like machine code, beneath the need for human understanding.
🌞 I'm curious if/how Alpöge's result will shift Terence Tao's view. The way I see it, we're about to witness a new generation of mathematicians who will treat Lean proofs like machine code, trusted because it checks out - not because it was read, checked, understood, and given a sign of approval by peer review.
🌑 Who knows, perhaps the volume of 'opaque' mathematics - math which we rely on but no one has read and understood - will start growing in scope, until we pretty much lose our complete grasp of the inner workings of mathematics.
We're living in exciting times
- that's for sure.
Math Major | Daily Diary D0059
🥱 Okay. I'm seriously tired.
This week I have been coding, doing math, and writing a whole lot. Often going to bed at 2 am, and back up at 7 am. Need some sleep, for sure.
🖊️ Anyway, one of the things I've been working on is a draft of an epistemology paper exploring a topic I started thinking (and writing) about this summer:
generative asymmetry.
🧫 Yes, I'm primarily talking about the challenges of Generative AI triggering a flood of results in mathematics - but it is my intention to try and extend the idea to (amd draw connections from) other domains that manifest the same pattern.
📬 The draft still needs a hell of a lot of work - but I figured I'd share it now, in order to break the fear of looking stupid. I hope that in a few months, the draft will be in a state where I feel comfortable asking someone to peer review it.
🔥 As soon as a I get some time over, I'll add a comment section to my website - with the hopes that someone might want to read the draft and suggest improvements.
🧗♂️ Anyway, gotta keep moving.
There are lots of fun things to work on.
https://t.co/SacPHMDCZ4
@xKaka1456 "I'm writing an essay on that topic. Using the LHC as an example of capital-intensive research slowly sunsetting the era of the lone genius seated at the base of an apple tree, doing frontier research."
Harvard mezunu Türk asıllı matematikçi Levent Alpöge, 167 yıldır çözülemeyen ve başına 1.000.000 dolar ödül koyulan Riemann Hipotezi’ni çözdüğünü açıkladı.
— Hipotez, asal sayıların tamamen rastgele olmadığını, aslında mümkün olan en düzenli biçimde dağıldığını ortaya koyuyor.
— Alpöge, yüz yılı aşkın süredir çözülemeyen problemi Claude yapay zekasını kullanarak çözdüğünü duyurdu.
📅 I'm going to get started here (see link), and spend the next 30 days exploring Lean and autoformalization.
🗒️ I'm also going to put up a notice on campus and ask if other math students want to get together twice a week and explore Lean and AI math workflows.
https://t.co/HYsXLhdhFo
Math Major | Daily Diary D0075
For anyone doing mathematics with Astra/Claude, and doing autoformalization of math papers using Lean
- how did you get started?
Appreciate any tips on tutorials and learning paths.
Thanks for suggesting that paper - seems it could be relevant for the following areas:
✅ Radar range equation
✅ Radar detection probability
✅ Radar coverage estimation
✅ Target detection
✅ Radar resolution
✅ Target tracking
✅ False alarms and noise
✅ Noise and interference
I'm also reading Poisel's books, especially his use of Grassmann for geoloc. The Target book has an entire appendix on Grassmann algebra.
@TweetsA91248@HarryStoltz1 "Contribute to the collective body of knowledge."
At a time when world class mathematicians and Fields Medalists are panicking about AI devouring their field; that's the best time to enter the field. Got it.
Few people will care - mainstream reaction will barely register. Think Hubble telescope-levels of excitement:
<excited astronomer>
"Folks! With the amazing Hubble telescope we can see the dead star Earendel as it was 13 billion years ago!!!"
<everyone else>
"Cool.
Anyway, I'll have a Big Mac, a large fries..."
Thank you for the kind and wise words.
Uni is funny because it always feels like all the exciting stuff is happening outside of class (no offense to my lecturers).
People are impatient - we want to play with fire in the form of the bleeding edge of technology. Sitting in class and learning the fundamentals of mathematics isn't always super exciting - but neither is hitting the gym and doing pushups - but strong fundamentals matter. 🏋️♂️
Just have to be patient, I guess. 😀
Wishing you a pleasant weekend!
Math Major | Daily Diary D0074
🐶 Working on a side project, but I'm not too happy with the result so far - I need to clean up lots of things and clarify others. But at least it's a start.
🐱 At the same time - I'm quite tired.
I want to focus exclusively on Uni work for the next six weeks (exams, labs, and lots of homework).
🐭 Anyway, RoboCheck is a side project that I'm working on during late evenings (nights) and early morning . It's nothing much, but I thought I'd share an update before I dial down the build-arc and focus on Uni.
🧀 In RoboCheck, a user can select requirements for a robot tank. These reqs will then be used to generate robocode. the robocode is verified for correctnes using the formal verification language Verus.
Finally, the user's robot combats other robots, in an arena, and the result generates suggestions on things to improve.
🔨 I'm working on adding a workflow where the user can freely write things they want the robot to do (combat strategies, movement, scanning, etc) but I'm not finished with that part.
🪛 It's been fun mixing natural language, formal verification, and robot combat tanks.
🪤 Further to follow (in a couple of months).
Math Major | Daily Diary D0073
Was about to hit my weekly Claude limit- five days early,
only to realize that the awesome folks at @AnthropicAI had blessed me with a free reset. 🥳
🙌 😀 Thank you @bcherny, and everyone over at Anthropic, for blessing developers like me with the most fulfilling developer experience I've ever had.
Cheers!
Math Major | Daily Diary D0072
🥳 I'm now a proud owner of a first edition 1962 book by E.W. Dijkstra.
🙌 Even though ALGOL is no longer relevant - reading the words of Dijsktra, one of the greats of computer science, never gets old.